Putting predictive analytics and medical device data to work reducing a common physician alert by 99%
Can predictive analytics, fueled by real-time medical device data, accurately identify patients at-risk for post-op respiratory depression without inducing alarm fatigue in clinicians?
That was the question Virtua Memorial Hospital in Mount Holly, New Jersey, and medical device integration and analytics company Bernoulli set out to answer when they embarked on a recently-concluded initiative.
Bernoulli's middleware and a rules-based analytics engine were deployed to continuously surveil the patients; it was not only able to able to identify at-risk patients, but reduced false alarms by more than 90 percent, the hospital reported.
Here’s a look at how they made it happen.
The problem: alert fatigue
According to the Joint Commission's Sentinel Event database, 29 percent of adverse events in hospitals are related to improper surveillance. The ability to track patients across care settings, continuously integrate new medical devices and distribute real-time patient monitoring to centralized dashboards and mobile devices could have a profound impact for hospitals seeking to create a foundation for addressing other hospital safety issues.
"As an anesthesiologist, I see a number of opioid-related side effects during the postoperative period, including respiratory depression," said Leah Baron, MD, chair of the department of anesthesiology at Virtua Memorial Hospital. "In our institution, a significant number of our surgical patients are diagnosed with obstructive sleep apnea or suspected to have obstructive sleep apnea. These patients are known to be more susceptible to opioid-induced respiratory depression."
By observing this subset of patients within the study in the recovery room and on med-surg units, caregivers recognized that better monitoring modalities were needed for early detection of respiratory depression. The goal was to create a safer environment for the unit's high-risk surgical population.
"The inability to identify respiratory depression early is not just a patient safety problem," said Baron. "Rescuing these patients is also costly in terms of resource utilization, morbidity and mortality. Finding ways to safely monitor high-risk surgical population on med-surg units and decreasing utilization of more expensive beds by this high-risk population could provide significant cost savings for the institution."
The research set out to determine if selectively delayed notifications using adjustable, multi-variable thresholds to identify clinically actionable events could significantly reduce the overall number of alarms without increasing risks to patients in a post-anesthesia care unit.
"The study measured pulse, oxygen saturation, respiratory rate, and end-tidal carbon dioxide continuously and compared alarms received through the bedside monitoring device with remote alerts designed to trigger only after a selective delay," according to John Zaleski, executive vice president and chief analytics officer at Bernoulli.
The goal, he said, was to "reduce the total number of alarms without increasing risk to patients who have been diagnosed with or are at risk for obstructive sleep apnea."
Using only sustained alarms as the filter for notifications reduced alerts from 22,812 to 13,000, a number high enough to still cause alarm fatigue. However, passing multiple series of data through a multi-variable rules engine that monitored the values of pulse, oxygen saturation, respiratory rate, and end-tidal carbon dioxide in order to determine which alarms to send to the nurse-call phone system brought the number of respiratory depression alerts down to 209 – a 99 percent reduction, the hospital reported.
More important, the analytics alerted for every patient that experienced an actual respiratory depression episode, the hospital added.
"The use of real-time, clinically actionable data requires more than filtering based on alarms and vital signs thresholds," said Zaleski. "Combining analysis with real-time data at the point of collection creates a powerful tool for prediction and decision support."
Continuous clinical surveillance allows clinical staff to monitor patients remotely. Networked laptop and desktop computers, as well as scrolling message bars, can provide clinical staff with access to data and alarms from all surveilled patients, he said. In addition, alarms can easily be routed to central stations via dashboards or mobile devices and pagers, he added.
Ultimately, caregivers in the study identified sustained combinatorial alarms that principally combined respiration rate and end-tidal carbon dioxide and that correlated with clinical significance when using the respiratory depression surveillance solution, Baron said. The most significant drop in hypopnea/hypoventilation combinatorial alarm occurred at 18-second alarm delay, reducing the number of alarms from more than 4,500 to 209, she added.
"An important observation made during this study was that remote alarm communication was an important adjunct to in-room monitoring alarm annunciation," Baron said.
"A key argument that is made for in-room annunciation in the case of conscious or waking sleep apnea patients is the room audible alert," she explained. " Yet, in every observed case of opioid-induced respiratory depression, the in-room audible annunciation had no effect on waking or stirring the patients. Hence, the need for a remote monitoring capability to catch such instances is motivated more strongly to ensure patients do not slip through the cracks."
Hospitals need to take a system-wide inventory of alarms, apply analytics to understand their value and convene project teams to embark on a technology transformation, said Zaleski.
"This transformation should include the perspectives of frontline clinicians and should respect the significance of disrupting workflows," he said. "Hospitals should employ smart technologies to ensure only actionable alarm signals are sent to clinical staff. The deployment of continuous clinical surveillance requires a careful investment of time and money in addition to workflow considerations to ensure the highest level of patient safety."
Big Data & Healthcare Analytics Forum
The San Francisco forum to focus on utilizing data to make a real impact on costs and care June 13-14.